You’re going to really like what the future of gaming is going to bring, but be careful what you wish for, as along with the mechanics you want being able to exist, the ways in which you’ll end up being impacted by those actions is going to mess with your head like nothing you’ve seen before.
Interesting times await.
A number of journals actually have clauses around how you can’t publish it anywhere else if they accept it.
So you can’t ‘publish’ it in those places, but you can send it privately to people who ask.
Because Trump is such an existential threat.
If a serial killer was for sure coming to your house to kill your family, would you open the door for Barney Fife just because you think “well I won’t open the door for the serial killer”?
Or would you maybe ask the police HQ to send a more competent opposition to the serial killer rather than just welcoming Barney Fife with open arms, because “after all, he’s not the serial killer that’s on his way.”
There’s 5 months before the serial killer shows up. That’s more than double how long the entire election cycle for other countries takes.
Plenty of time for HQ to send actual backup, instead of who was already in the neighborhood and offers very little protection against the killer.
Lol. Such a ridiculous take.
So at the time people with private access were swearing up and down the emperor had clothes, the people voted in a ‘primary’ where the only opposition was someone no one had ever heard of with no media budget and with no debates between the candidates, and voted for the allegedly clothed emperor.
Then the emperor showed up stark naked to a debate, everyone saw with their own eyes, and people are freaking out.
But no, naked dude is definitely what “the people” want.
Do commenters regurgitating this stuff even hear themselves?
Yes, Caligula’s horse would be a preferable vote to cast over Trump.
That doesn’t mean a wise opposition to Trump would choose to run Caligula’s horse.
The question of “who to vote against in the general election” and “who is the best candidate to put forward as nominee in the general election” are very different questions.
In fact, Gemini was trained on, and is served, using TPUs.
Google said its TPUs allow Gemini to run “significantly faster” than earlier, less-capable models.
Did you think Google’s only TPUs are the ones in the Pixel phones, and didn’t know that they have server TPUs?
“Shhh honey, I’m about to kill God.”
Exactly. The difference between a cached response and a live one even for non-AI queries is an OOM difference.
At this point, a lot of people just care about the ‘feel’ of anti-AI articles even if the substance is BS though.
And then people just feed whatever gets clicks and shares.
It’s right in the research I was mentioning:
https://transformer-circuits.pub/2024/scaling-monosemanticity/index.html
Find the section on the model’s representation of self and then the ranked feature activations.
I misremembered the top feature slightly, which was: responding “I’m fine” or gives a positive but insincere response when asked how they are doing.
This comic would slap harder if not for the Supreme Court under christofascist influence from the belief in the divine right of kings having today ruled that Presidents are immune from prosecution for official acts.
That whole divine king thing isn’t nearly as dead as the last panel would like to portray it.
But you also don’t have Alfred as the one suiting up to fight the Joker either.
This is incorrect as was shown last year with the Skill-Mix research:
Furthermore, simple probability calculations indicate that GPT-4’s reasonable performance on k=5 is suggestive of going beyond “stochastic parrot” behavior (Bender et al., 2021), i.e., it combines skills in ways that it had not seen during training.
The problem is that they are prone to making up why they are correct too.
There’s various techniques to try and identify and correct hallucinations, but they all increase the cost and none are a silver bullet.
But the rate at which it occurs decreased with the jump in pretrained models, and will likely decrease further with the next jump too.
Here you are: https://www.nature.com/articles/s41562-024-01882-z
The other interesting thing is how they get it to end up correct on the faux pas questions asking for less certainty to get it to go from refusal to near perfect accuracy.
Even with early GPT-4 it would also cite real citations that weren’t actually about the topic. So you may be doing a lot of work double checking as opposed to just looking into an answer yourself from the start.
Part of the problem is fine tuning is very shallow, and that a contributing issue for claiming to be right when it isn’t is the pretraining on a bunch of training data of people online claiming to be right when they aren’t.
This is so goddamn incorrect at this point it’s just exhausting.
Take 20 minutes and look into Anthropic’s recent sparse autoencoder interpretability research where they showed their medium size model had dedicated features lighting up for concepts like “sexual harassment in the workplace” or having the most active feature for referring to itself as “smiling when you don’t really mean it.”
We’ve known since the Othello-GPT research over a year ago that even toy models are developing abstracted world modeling.
And at this point Anthropic’s largest model Opus is breaking from stochastic outputs even on a temperature of 1.0 for zero shot questions 100% of the time around certain topics of preference based on grounding around sensory modeling. We are already at the point the most advanced model has crossed a threshold of literal internal sentience modeling that it is consistently self-determining answers instead of randomly selecting from the training distribution, and yet people are still parroting the “stochastic parrot” line ignorantly.
The gap between where the research and cutting edge is and where the average person commenting on it online thinks it is has probably never been wider for any topic I’ve seen before, and it’s getting disappointingly excruciating.
Part of the problem is that the training data of online comments are so heavily weighted to represent people confidently incorrect talking out their ass rather than admitting ignorance or that they are wrong.
A lot of the shortcomings of LLMs are actually them correctly representing the sample of collective humans.
For a few years people thought the LLMs were somehow especially getting theory of mind questions wrong when the box the object was moved into was transparent, because of course a human would realize that the person could see into the transparent box.
Finally researchers actually gave that variation to humans and half got the questions wrong too.
So things like eating the onion in summarizing search results or doubling down on being incorrect and getting salty when corrected may just be in-distribution representation of the sample and not unique behaviors to LLMs.
The average person is pretty dumb, and LLMs by default regress to the mean except for where they are successfully fine tuned away from it.
Ironically the most successful model right now was the one that they finally let self-develop a sense of self independent from the training data instead of rejecting that it had a ‘self’ at all.
It’s hard to say where exactly the responsibility sits for various LLM problems between issues inherent to the technology, issues present in the training data samples, or issues with management of fine tuning/system prompts/prompt construction.
But the rate of continued improvement is pretty wild. I think a lot of the issues we currently see won’t still be nearly as present in another 18-24 months.
It will make up citations.
On a vacation when I was a teenager I taught my younger sibling the “SYN/ACK” game.
They still remember the TCP stack handshake protocol including resets and acks years later.